Skip to content
#Research
Nov 2Mon
  1. arXiv cs.RO62

    Nexus-VLA: Connecting Human and Robot Experience with a Shared Hierarchical Semantic Latent Space

    The authors propose Nexus-VLA, a hybrid policy learning framework that enables cross-embodiment knowledge transfer through end-to-end human-robot joint training, allowing human manipulation experience to directly contribute to robot policy learning. The framework first establishes hierarchical cross-embodiment semantic alignment from task intent to fine-grained interactions, then encodes multi-level shared knowledge into a shared manipulation latent representation used directly for continuous action generation. In RoboCasa-GR1, OOD settings, and real-robot experiments, Nexus-VLA achieved an 80.5% task success rate on RoboCasa-GR1.

  2. arXiv cs.RO58

    Suspended fabric robotic manipulation surfaces: modeling, design space, and multimodule extension

    This paper proposes a separable catenary model for suspended-fabric Robotic Manipulation Surfaces (RMS) that can quickly compute the surface shape for given actuator heights, as well as the actuator heights needed to achieve a desired local slope. The closed-form inverse solution lets a simple PID controller position objects on the fabric surface, and the model relates fabric length, module span, and actuator stroke to the reachable workspace, yielding a lower bound on minimum fabric length and design charts.

  3. arXiv cs.RO62

    Open-Source Knee Exoskeleton: Generating Biological Torque Control with Musculoskeletal Simulation

    A research team has released an open-source knee exoskeleton built from off-the-shelf components, whose biological torque controller is generated entirely by simulation, requiring no training data collected on the target device. The method uses a muscle-driven imitation policy to reproduce squatting motions from a public biomechanics dataset, then distills the knee joint torque into an estimator that relies only on onboard sensors.

  4. arXiv cs.RO51

    Language-conditioned behavioral cloning for autonomous endoscope guidance

    The study proposes language-conditioned behavioral cloning for autonomous endoscope guidance, defining 6 clinically relevant guidance behaviors and collecting 235 subtask demonstrations. In preliminary phantom experiments, a single visuomotor policy achieved 100% success on focus, overview, and stop, 80% on search-tip and follow-tip, and 50% on horizon leveling. The authors note that the failure modes indicate more complex guidance tasks require temporal context.

  5. arXiv cs.RO43

    Adaptive Coverage Algorithms for Mobile Ad Hoc Networks in Dynamic Worlds: CADENCE and DADENCE Extended to DPOCGAGP

    The study extends two algorithms, CADENCE and its decentralized version DADENCE, to the dynamic version of POCGAGP (DPOCGAGP), adding two deployment-time perturbation settings, Agent Loss and World Change, for multi-agent systems to establish a fully covering communication network in unknown constrained spaces. On more than 4,500 benchmark test cases of varying scale and structure, the improved algorithms outperform another state-of-the-art benchmark algorithm. The work focuses on simulation benchmark evaluation and has not yet involved real-robot validation.

  6. arXiv cs.RO64

    Online Learning of Heterogeneous Locomotion Dynamics for Precise Mobile Manipulation

    The study proposes an online method for learning closed-loop dynamics under a frozen locomotion policy, using a contaminated Gaussian model to distinguish predictable from unpredictable responses and combining a Gaussian process with random Fourier features to build an online learner that reduces bias from heterogeneous observations. Combined with model predictive control (MPC), the method predicts floating-base motion more accurately; in simulation and hardware experiments, end-effector tracking error was 0.01 m, and a finite-sample error bound for the online learning is provided.

  7. arXiv cs.RO80

    Movens-Zero: Scaling humanoid motion foundation models with large-scale human videos

    The research team proposes Movens-Zero, a scalable understanding-generation-execution architecture that supports real-time closed-loop humanoid motion generation under text, speech, music, and video inputs. Its accompanying dataset, Movens-5M, is built from Internet videos through clip generation, quality filtering, semantic annotation, motion recovery, and physical refinement with humanoid retargeting. It contains 15 million human motion sequences and 5 million physically executable humanoid motion sequences, totaling over 10,000 hours.

    Why it matters: This technical report presents a complete path for building a humanoid motion dataset from Internet videos and training a unified understanding-generation-execution model, offering a reference for data scale and architecture design.

  8. arXiv cs.RO80

    SimuGuard: Auditing Contact Artifacts in Simulation Benchmarks Such as LIBERO and RoboTwin

    The authors propose SimuGuard, an auditing plugin for LIBERO, RoboTwin, RoboCasa, and RoboDojo that identifies failures caused by the simulator rather than the policy. On RoboTwin, the authors trace these failures to the physics engine's collision detection: when geometry is 1 to 2 centimeters apart, the engine reports a penetration of about 2 centimeters, and the solver ejects the object at several meters per second because the distance algorithm discards triangles below a fixed area threshold.

    Why it matters: The original traces failures in simulation benchmarks to a physics engine collision detection constant and provides a before-and-after success rate comparison, which can serve as a reference for teams doing simulation evaluation.

  9. arXiv cs.RO67

    SketchMate: Building Sketch Models with VLMs for Zero-Shot Precision Insertion

    SketchMate has a vision-language model build a sketch model for insertion tasks, consisting of a few geometric primitives and two mating coordinate frames, which a planner and compliant controller then execute. In a strict kinematic test of 69 simulated insertion episodes, the method inserted 88% of parts at the assembled pose, while ReKep and MOKA using the same VLM achieved no successful insertions.

  10. arXiv cs.RO38

    Integrating Low-Thrust Transfer Dynamics into POMDP Planning for On-Orbit Servicing Missions

    This research models low-thrust (LT) orbital transfers within a partially observable Markov decision process (POMDP) for on-orbit servicing (OOS) mission planning, adding the servicing spacecraft's remaining propellant to the state space to support mass updates. High-thrust transfers are modeled with a precomputed transfer matrix, while low-thrust transfers are represented by an analytical GEO phase model, with propellant consumption and maneuver feasibility computed for given transfer-duration alternatives. In a simplified OOS scenario, SARSOP is used to solve and compare high-thrust and low-thrust policies, quantitatively assessing the feasibility and operational impact of the low-thrust option.

  11. arXiv cs.RO58

    Visual-Inertial-Encoder RGB-D SLAM for Differential Wheeled Robots

    The study proposes a generative measurement model for encoders and derives the maximum a posteriori estimation expression and the Jacobians required for optimization, building a tightly coupled visual-inertial-encoder RGB-D SLAM system. On a self-built dataset, the system's real-time state estimation achieves a root mean square error of about 2-7 cm, used to compensate for the failure of pure vision solutions during temporary feature point loss and fast motion. The source code and the dataset containing encoder information have been made public.

  12. arXiv cs.RO67

    Higher-Order Kinematic Derivative Methods for Aspect-Preserving Real-Time Numerical Inverse Kinematics

    The paper uses higher-order kinematic derivatives of serial robots to propose an aspect-preserving real-time numerical inverse kinematics method, yielding the QuIK4 and QuIK5 solvers with fourth- and fifth-order convergence. The method also includes an aspect-preserving step limiter that extrapolates the Jacobian determinant along the step to predict and avoid singularities, as well as a certifier that uses Hermite interpolation to verify the aspect of solutions and a certified restart algorithm. The authors claim the failure rate is reduced by up to four orders of magnitude compared with existing methods, and they release a C++ library with ROS 2 integration.

  13. arXiv cs.RO36

    When Architectural Spaces Begin to Act: How Architectural Robots Reshape Living and Spatial Control

    Architectural robots, such as robotic partitions, ceilings, furniture, and facades, can move autonomously and reconfigure space. The author reflects on this: their impact goes beyond the moment of motion to include the spatial configurations that persist after motion stops. Such actions may benefit some occupants while disadvantaging others, and multiple robots together can make an entire architectural space appear to act. The author argues that understanding robotic architecture requires looking beyond short-term human-robot interaction to how people inhabit, negotiate, and retain control over autonomously reconfigurable spaces.

  14. arXiv cs.RO60

    Flexible Sensing Robot Manipulation: Deformation Estimation, Kinematics, and Control Based on Distributed Inertial Measurement

    This paper proposes a flexible sensing manipulation framework that directly connects the structural dynamics of flexible links with sensing, kinematics, and control. The method first identifies the dominant vibration modes of each flexible link according to response convergence and excitation bandwidth criteria. Under the condition of full-rank modal observability, m retained modes require only m independent slope measurements and one reference IMU, and the continuous link deformation is reconstructed accordingly for flexible sensing forward kinematics and a PDE-based controller.

  16. arXiv cs.RO58

    Study: Biomechanical metrics cannot predict perceptual and cognitive consequences of force augmentation

    An arXiv study used an elbow-assist exosuit to examine the physiological, perceptual, and cognitive consequences of force augmentation during lifting, finding that although assistance reduced muscle activity, behavioral outcomes did not simply correspond to this biomechanical change. Assistance reduced perceived weight, but this effect was strongly shaped by participants' initial experience; assistance also selectively reduced adaptation to changes in environmental conditions, and differences in effort-based decisions were likewise shaped by initial experience rather than simply following the decline in muscle activity or perceived fatigue. The authors therefore argue that the perceptual and cognitive consequences of force augmentation cannot be inferred from biomechanical benefits alone and must be assessed alongside physical performance.

  17. arXiv cs.RO60

    Interaction-Aware Sampling-Based Motion Planning for Autonomous Surface Vessels in Urban Waterways

    The authors propose a decentralized motion planner based on model predictive path integral control that jointly computes the ego vessel's trajectory and predictions of other vessels' trajectories, encoding traffic rules as discontinuous costs without requiring smooth approximations. In maps taken from real sections of Amsterdam's canals, 780 randomized runs all reached their goals without collisions or deadlocks, outperforming an optimization-based planner designed for the same domain as well as a version of the proposed planner that does not account for interaction. Two scaled vessel prototypes completed head-on and crossing encounters without rule violations, planning and localizing onboard at 10 Hz in an outdoor environment.

  18. arXiv cs.RO62

    OpenVIO: An Open-Source Visual-Inertial Odometry System

    OpenVIO is a visual-inertial odometry (VIO) system released as open source under the MIT license. It supports pinhole and fisheye cameras and extends to stitched 360° images. The authors describe the design considerations for each module of the system and benchmark it against other open-source VIO systems on public datasets. The code is open source, and the paper is 7 pages with 6 figures.

  19. arXiv cs.RO58

    GPU-Accelerated Incremental Euclidean Distance Transform for Online Motion Planning of Mobile Robots

    This paper proposes a voxel mapping system that uses parallel computing to simultaneously generate occupancy grid maps (OGMs) and Euclidean distance transforms (EDTs), and incrementally builds a global EDT to output high-frequency local distance information for online motion planning. The system uses GPU programming to compute EDT in parallel within local volumes, then continuously integrates new observations into the global EDT via a parallel wavefront algorithm while preserving historical observations.

  20. arXiv cs.RO62

    ResMerge: Residual Policy Learning and Merging for Continual Adaptation of Pre-trained Robot Policies

    ResMerge proposes a two-stage framework that learns mergeable residual policies atop a frozen pre-trained policy: residual reinforcement learning obtains task-specific corrections during adaptation without additional expert demonstrations, while parameter-space merging consolidates compatible residuals to reduce stored parameters.

  21. arXiv cs.RO78

    ∇Mimic: Efficient Whole-Body Control for Humanoid Robots with Differentiable Simulation

    The authors propose ∇Mimic, which uses exact gradients from a differentiable simulator to train whole-body motion-tracking policies for humanoid robots, reaching a score of 0.82 in 20 minutes and 0.874 in 1.8 hours on a single GPU, whereas the world-model method takes 1.2 hours to reach 0.82.

    Why it matters: The paper presents a complete scheme for training humanoid policies with exact gradients under differentiable simulation and open-sources the engine and code, making it easy to compare the efficiency of first-order and zeroth-order methods.

  23. arXiv cs.RO66

    RTPolicy: Real-time multi-view visuomotor policy for edge devices, inference at 32.9 Hz

    RTPolicy is a multi-view visuomotor policy that uses view-aware feature fusion and visual compression to aggregate multi-camera information into a compact token representation, and uses a flow-matching action generator to reduce action-generation overhead. On the 50 tasks of the RoboTwin 2.0 single-task clean evaluation, it achieved an average success rate of 52.26%, higher than the strongest leaderboard baseline π0 at 46.4%.

  24. arXiv cs.RO62

    VLA policies still start executing without instructions: research on unintended task execution and start suppression

    The study applied five types of instructions—empty, neutral, meaningless, infeasible, and wrong-task—to six model checkpoints in the LIBERO-Goal simulation, holding the initial state fixed and varying only the instruction, to examine whether the VLA policy starts, which task it completes, and what the instruction determines.

  25. arXiv cs.RO61

    Study provides complete analytical characterization of optimal collision-free motion for two robots

    The paper derives a complete analytical characterization of optimal collision-free trajectories for two robots moving from rest to rest under double-integrator dynamics and a squared-acceleration objective. It proves that these trajectories belong to eight contact families and gives the family determination for each combination of initial and final positions. The authors analyze the geometric properties of the optimal trajectories, provide a numerical computation algorithm, and discuss how these results can improve the performance of existing numerical trajectory optimizers.

  26. arXiv cs.RO67

    SCI: A Contact-Based Optimization Method for Collision-Free Motion Planning for a Million Robots

    The authors propose the sequential contact insertion (SCI) algorithm, which organizes optimization around contact points where robots reach the minimum allowed separation and analytically reconstructs trajectories between contacts. In numerical experiments, the method can plan for one million robots within minutes, while thousands of robots take less than one second; on problems solvable by a fixed-grid sequential convex programming baseline, SCI achieves median speedups of several orders of magnitude with comparable solution accuracy.

  27. arXiv cs.RO58

    AC-SMPC: Differentiable Actor-Critic MPC with State Constraints for Safe and Fast Robot Navigation

    The authors propose AC-SMPC, an actor-critic model predictive control framework with state constraints for safe navigation in constrained dynamic environments. The method represents the robot and obstacles as vertices and the workspace as half-spaces, locally linearizes the nonlinear dynamics, and convexifies hyperplane-based collision avoidance constraints into a differentiable MPC layer; the actor adjusts MPC objective parameters via closed-loop feedback, while the critic evaluates long-term task performance.

  28. arXiv cs.RO66

    SAFARI: Semantic- and Frontier-Driven Autonomous Risk Inspection Framework

    SAFARI treats object detection as a native channel of a unified probabilistic map, letting geometric and semantic evidence jointly drive exploration and planning. The authors completed 11 missions on a quadruped platform across three field sites, achieving an average coverage of 98.7% of the accessible mission area, with 97.6% of mission time requiring no operator intervention, and detecting objects with characteristic dimensions as small as 15 cm. Ablation experiments show that object-size awareness and safety-aware path planning are critical for hazardous-object search.

  29. arXiv cs.RO58

    Physically Constrained Kinematic Calibration for High-Precision Robotic Needle Guidance

    The paper proposes a bound-constrained kinematic calibration method to improve the positioning accuracy of a 6-DoF robotic needle-guidance manipulator. The method constrains a 19-parameter geometric identification model with physically plausible parameter ranges, treating the bounds as active optimization constraints to mitigate parameter coupling and overfitting. In simulation, end-effector RMSE dropped from 3.422 mm to 0.690 mm (a 79.8% improvement); in benchtop experiments, trajectory validation RMSE was 0.865 mm and coordinate grid validation RMSE was 0.610 mm.

  30. arXiv cs.RO69

    BioAgent: An embodied agent that autonomously acquires skills and expands capabilities for long-horizon laboratory protocols

    BioAgent is an embodied agent for long-horizon laboratory protocols that acquires new skills by matching human-reviewed requirements with validated skills, selecting tool-based implementations for missing behaviors, and requesting human demonstrations as needed. On the second task, it reused 17 skills unchanged, modified 2 existing implementations, and added 5 new skill types; on two dual-arm laboratory tasks with 20 and 31 ordered skill invocations respectively, it achieved complete-task success rates of 66% and 58%, with each task run 50 times and autonomous recovery enabled.

  34. arXiv cs.RO56

    TaSP-KKT: Geometry-Aware Humanoid Motion Retargeting Based on Manifold ODEs

    The study proposes TaSP-KKT, which models the continuous-time limit of OmniRetarget's per-frame KKT conditions as an ODE on the robot configuration manifold and implements motion retargeting through forward discretization. The method requires only one weighted tangent-cone projection and a closed-form manifold update per step, avoiding repeated inner SQP iterations. In numerical experiments, retargeting quality is comparable to OmniRetarget but slightly lower, while running speed is about twice as fast.

  35. arXiv cs.RO60

    Joint-Health-Aware Reinforcement Learning Framework for Robust Control Under Actuator Degradation

    The study proposes a joint-health-aware reinforcement learning framework that encodes actuator condition in a continuous joint-health vector and uses HealthNet to estimate each joint's health from recent observations and actions, allowing the policy to redistribute control effort across healthy and degraded joints.

  36. arXiv cs.RO62

    HandMimicX: Learning Object-Centric Dexterous Manipulation Policies from Egocentric Videos

    HandMimicX proposes an object-centric framework for learning dexterous manipulation skills from egocentric human RGB-D videos. The method defines the reinforcement learning objective in terms of object motion, decoupling skills from specific hand trajectories, and imposes constraints on hand-object interaction states during grasping and manipulation to obtain more anatomically plausible hand configurations. Simulation and real-robot experiments show that the method learns efficiently from human videos and transfers skills to novel spatial configurations.

  37. arXiv cs.RO68

    HIPC: Humanoid Iterative Planning and Control via Goal-Conditioned Motion Closure

    The authors propose HIPC, a real-time humanoid iterative planning and control framework that couples a generative model with motion tracking into a unified planning-and-control paradigm. HIPC uses goal-conditioned motion planning and trains with a canonical motion representation to avoid gimbal-lock singularities under extreme poses, while mitigating out-of-distribution shifts caused by real-robot state feedback through two strategies: deviation-decayed recovery and history re-anchoring. Experiments show that HIPC achieves the best goal-reaching accuracy among existing methods and exhibits robustness to physical disturbances and the ability to recover after falls.

  38. arXiv cs.RO67

    Hourglass: Cross-Embodiment Motion Planning by Coupling a Body-Agnostic World Model with Differentiable Ego-Dynamics

    The paper proposes the Hourglass architecture, which decomposes cross-embodiment motion planning into a body-agnostic world model and each robot's differentiable ego-dynamics template, coupled only through a camera ego-motion interface. On seven simulated robots across five motion types, this generalist policy achieves success rates 3.8–10.6 percentage points higher than imitation learning and diffusion policy baselines on the same world model, matches each robot's specialized policy without dedicated training, and comes within 1.4–2.8 percentage points of a parameter-conditioned MPPI teacher.

  39. arXiv cs.RO58

    LEXP: Learning Robotic Manipulation from Physical Experience via Context-Space Adaptation

    The authors propose LEXP, a framework for learning from physical experience through context-space adaptation that combines task-level planning with an experience module, reusing execution experience to generate contextual feedback while model weights and the action interface remain fixed. Across 12 real-world manipulation tasks, LEXP achieved 23/36 successes (63.9%), compared with 19/36 (52.8%) using memory alone and 2/36 (5.6%) for vanilla ChatGPT-6-Astra.

  40. arXiv cs.RO59

    TRACE: Target-State Risk-Gated Execution for Embodied Visual Tracking

    The research team proposes TRACE, an execution-time target-state risk gate that intervenes selectively while keeping the learned policy unchanged. They first used PPO to update only the planning head of a 0.6B VLA tracker; success and following rates improved across all three EVT-Bench settings, but the collision rate rose in distracted tracking scenarios.